Paul Krugman: Nobel Prize or Academy Award? When economic theory is a tower of babel

Below is my response to Krugman’s comments in defence of new trade theory. It’s not generated any discussion on the Mandarin or Evonomics, but perhaps it will here. Apologies for the delay in getting it onto Troppo – I’ve been travelling.

I recently criticised contemporary economics in a speech launching Max Corden’s memoirs. Economic theory threatens to become a Tower of Babel, preoccupied with the world within its models and irrelevant to policy. This has crowded out an older style of which Corden was an exemplar, in which economics aspired to be the supremely useful social science, policy-relevant as clarified commonsense.

I cited Paul Krugman’s claim that macroeconomics had actually regressed as a result, but I also pointed out that Krugman had himself been the architect of the same kind of regression. He won a Nobel Memorial Prize for his trouble.

In Krugman’s characteristically vigorous and clear-headed response to my speech, he suggests that this is the ‘money quote’ from my speech.

Krugman 1 about the most brilliant and useful economist we have. But his most brilliant work wasn’t useful, and his most useful work isn’t brilliant”

We’ll get to what I think was the money quote, but first … Krugman and I are in violent agreement about the subject of his response – the differences between the Macro and the Trade Towers of Babel.

In macro, the grand prize was ‘micro-foundations’ in which micro and macroeconomics would be unified into a crystalline structure, not unlike Euclid’s Elements, as macro was rebuilt from an aggregation of perfectly rational optimising agents. As I mentioned, “this mindset gave additional licence to the motivated reasoning of the libertarian right”.

For many, that was the idea all along. Like the misdirection of a master magician, new classical models assumed away all the ‘imperfections’ that stare us in the face and make the macroeconomy such an ornery beast, both to live in and to understand; its frictions in transmitting information and incentives, its ‘animal spirits’ and tendency to self-fulfilling prophecies of gloom and euphoria, its ‘sticky’ prices and wages.

And voila! Unemployment disappears! The great depression was really millions of workers taking a spontaneous holiday. And forget government action to fight the business cycle. It’s misguided, and futile if not actively destabilising. And no, I’m not kidding.2

By my lights, new classical macro is the most egregious example, but the pattern recurs in field after field.3 Including Krugman’s new trade theory. The new models produced a Tower of Babel in which strong, robust predictions or policy conclusions were hard to come by. But, as Krugman replies, here the motive wasn’t to banish aspects of reality that were inconvenient for the modelling (or one’s ideology), but rather to invite them in.

What could possibly be wrong with that?


When the assumptions behind your model are simple and seem to formally capture the essence of something important, it can open up a world of insight. This occurred in the first wave of formalism in economics from around 1870 to World War II. At this stage, it’s easy to think that economics’ job is to simply expand this formal method, forever relaxing assumptions towards greater realism.

That’s what the brilliant and useful Paul Samuelson thought as he pioneered the new, mathematical economics from the late 1940s on, seeking to “accumulate a convergent body of econometric findings convergent on a testable truth”. Forty years on he lamented. “My confession is that this expectation has not worked out”.

We’ve been here before. In fact, Isaac Newton ran into the same problem. The handful of simplifying assumptions he made in his celestial mechanics give us a God-like vision of the heavens that’s intoxicated us ever since. But it turns out that its power degrades quite quickly once more than just two celestial bodies (like planets) enter the analysis.4

The early progress of formalism brought forth a vision that was almost as miraculous as Newton’s system and in some ways more intoxicating.5 Quite standard demand and supply diagrams give you a way to look at the chaos of economic life of such easy power that it feels you’ve got magic goggles. Suddenly, you can see how all manner of policies affect both the economy as a whole and the relevant groups as well. Rent control, tariffs, tax and welfare changes – you name it.

But everything has its price. The price of this vision is a long list of unrealistic assumptions. And a central assumption is that competition determines prices, not firms. And the fact is that, in all markets that aren’t for commodities, scale economies limit the number of competitors and give firms some pricing power – as Apple has over smartphones and McDonalds or even the local milk bar has over its burgers. Now there’s a serpent in paradise.

In my original speech, I quoted Max Corden’s mentor, J.R. Hicks, wrestling with the serpent on our behalf. He was trying to hang onto those goggles. Firms’ pervasive pricing power is modern economics’ ‘three body problem’. As Hicks points out, if firms have some pricing power, you don’t even know if they’ll increase supply when demand rises. They might just raise prices, or even cutsupply to boost prices further. And if your answer to any important question is “it all depends on the circumstances”, you don’t have much of a theory.

Hicks plumps for perfect competition because, although imperfect competition is more realistic, if we “abandon … the assumption of perfect competition … the basis on which economic laws can be constructed is … shorn away”. He concludes:

We must be aware, however, that we are taking a dangerous step, and probably limiting to a serious extent the problems with which our subsequent analysis will be fitted to deal. Personally, however, I doubt if most of the problems we shall have to exclude for this reason are capable of much useful analysis by the methods of economic theory..

Hicks’ advice isn’t to ignore scale economies but rather to take them into account either informally, or with ad hoc models fitted to specific policy issues and circumstances, just as a person with a map understands that they might have to work out the detail that it doesn’t cover.


Economics had changed by the late 1970s. The Dixit-Stiglitz model of imperfect competition and variants of it were becoming workhorses in various areas. So it made sense to deploy them in trade theory to see what insights they might turn up. But those models themselves were full of ad hoc assumptions – leaving Hicks’ wider concerns unaddressed,6 limiting the practical relevance of the modelling results. The culture of economic discourse had changed too. With academic specialisation proceeding apace, Hicks’ style of close, discursive reasoning became rarer and, perhaps more to the point, unnecessary to get journal articles accepted.

Max Planck told us that science progresses one funeral at a time. But that’s it regresses too. Like lots of intellectual innovators, Krugman had difficulty publishing his early new trade papers. Referees were unconvinced that an “understanding of these issues would be helped by writing down formal models”.7 Of course the papers deserved publication, but the referees’ concerns weren’t only valid, they were vindicated. In the upshot, as Krugman admits, this journey towards more ‘realism’ in formal trade theory, wasn’t very useful. It showed that:

  • scale economies give us more gains to trade (suggesting that traditional models can substantially underestimate the case for free trade) but
  • at the same time the models generate a plethora of ‘special cases’ enabling countries to advantage themselves with trade interventions to advantage certain firms.

And we already knew that. As trade theorist Ronald Wonnacott protested after a decade of new trade theory, it “may not yet have even scratched the surface. To identify one of the large number of possible new cases, all that’s required is the combining of one of the many causes of market failure with one of the many mercantilist instruments that could then be used to increase welfare”.

Here’s Krugman’s Nobel acceptance speech:

You may think all this is obvious, and it is – now. But it was totally not obvious before 1980 or so – except for some prescient quotes from Paul Samuelson, you really can’t find anyone describing trade this way until after the theory had been laid out in mathematical models. The plain English version came later.

And you should bear in mind that economists have been thinking and writing about international trade for a couple of centuries; to come along and say, “Hey, we’ve been missing half the story” was a pretty big thing.

Here we get to the nub of my beef with Krugman. Certainly there must be a division of intellectual labour to get anywhere. But formal theory should develop in close dialogue with – indeed to the extent possible should be enveloped by – close discursive reasoning. And theorists must respect the knowledge of others working in areas of relevance. My reaction – I think, were he alive, J.R. Hicks’ reaction – to the way trade theorists ignored scale economies before Krugman modelled them is “shame on them”, or more constructively, “how can we improve the disciplinary incentives that reward this groupish myopia?”.

Get Evonomics in your inbox

For all his brilliance, I’m repeatedly shocked at Krugman’s complacency on this score from his perch in the heights of the academy. As I put it in my speech, Krugman thinks that that “those who smell a rat at the dysfunctionality of all this should just get over themselves”. But it gets worse. What could Krugman possibly mean when he tells his Stockholm audience that “the plain English version” came after his theory? What’s his response to the list I mentioned in my speech?

  • Hicks (1959),
  • Linder (1961),
  • Elkan (1965),
  • Vernon (1966),
  • Balassa (1966),
  • Drysdale (1969),
  • Grubel (1967),
  • Corden (1970),
  • Lloyd (1975)
  • Gray (1976),
  • Bhagwati (1973),
  • Krueger (1978).8

I’m genuinely baffled, given Krugman would be familiar with most if not all these authors. In any event, if he wants a ‘money quote’ from me to respond to, that would be my pick.

All this while Krugman laments new classical macro’s bred-in insularity from those pesky facts in the real world.


Theory is a moon buggy for exploring terrain that’s difficult or impossible to explore any other way. We need our wits about us to build and to guide the buggy. The real distinction isn’t between formal and informal methods – they both need each other.9 It’s between the use of those methods that is skilful and fruitful, rather than just clever, and use that is less so – between good analysis (formal and discursive) and not-so-good analysis.

Formal models will be worthwhile if they raise logical possibilities leading to conclusions that are important and surprising and weren’t otherwise clear. Another kind of usefulness arises when they give us another, better understanding of the mechanisms causing the economic phenomena we observe.

To get back to what Krugman took to be my ‘money quote’, since he’s stopped using his brilliance to rediscover the obvious, his work has focused on real and pressing problems. I think of Krugman’s Nobel for his new trade theory as an Academy Award. He deserves the Nobel for calling the risks of a lost decade long before the reality was upon us and for his economic journalism – the best since Keynes. Though these days he occasionally apologises for the provisional nature of his sketched models on his blog for their falling short of the standards expected in academic publication, their usefulness shines through. So much the worse for academic publications which, in the age of the internet, we’ve made lamentably little progress in reinventing.10

While the sales pitch was always that new trade theory responded to developments in the world, its real focus was itself. In 1985, Brander and Spencer showed that an export subsidy could advantage a nation hosting one of the two firms in a duopoly where the other was located abroad – think Airbus and Boeing.11 Apparently, this “idea” generated a “flurry of excitement”.12It wasn’t wrong theory. It was bad theory, labouring to rediscover the obvious by imposing crude assumptions to get the maths out. The next year Eaton and Grossman published an analogous model with a slightly different (though equally crude) assumption about how the firms interacted.13 This reversed the result.14 Which paper was right? Neither. They both follow long-established conventions in the assumptions and the maths they use. So they look serious and, as we saw, they’re publishable. But, as a way of thinking about the difficult question of Airbus and Boeing’s rivalry, both tell us next to nothing. They’re silly; ‘academic’ in the bad sense.

In this sense, I think the heirs to Corden’s work in trade are those who’ve done the intellectual work to understand the patterns of trade and development we observe and to infer robust, or at least promising and insightful policy conclusions from them: people like Dani Rodrik and Ricardo Hausmann. They’re helping break open another citadel of groupish myopia – the Washington Consensus. They use theory of course, but not merely to rediscover the obvious and then reprove others’ for failing to render their thoughts visible to the high priests.

Postscript: The podcast of the article

2018 July 8


2 As Paul Romer puts it “Macro models now use incredible identifying assumptions to reach bewildering conclusions”.

17 To quote two emails I received from practicing economists – one in the markets and one in an international governmental organisation – in the upshot of this debate

18our aphorism was a good one. … 19t is, in a way, a marvellously pithy summary of the the whole of post-1980 economics – macro, monetary, trade (I assume), micro… the lot. When brilliant, a bit useless – when useful, all too often a statement of the bleeding obvious.”

20 could easily be describing aspects of health economics”.

4 Patterns that emerge robustly with two bodies often break down into non-recurring patterns bogging the analysis down into endless special cases some of which break down into chaotic motion. Though there are multiple bodies in the solar system, stability is assured by the fact that virtually all the relationships exist as one or more small bodies circling a much larger body so the third-body interactions have very marginal effects on the main pair. The planets perturb each others’ orbit infinitesimally.

The presence of three bodies does not produce chaos automatically, but raises the spectre of the same. See this essay for further elaboration. The point is simply that the capacity to understand even a very simple system can break down quite quickly once we move towards greater ‘realism’ or fidelity to the empirical world we can see.

5 It gives us a powerful new way of looking at the world. As Australia’s then Opposition leader, Professor of Business, John Hewson observed in 1991: “As soon as you get an equilibrium approach to life, suddenly you realise that a lot of what you’d thought was wrong”. But it threatens to be more intoxicating because, in addition to painting a picture of the way the world really is (if only the world could stop being so messy) it also offers a powerful vision of an ideal. When an anomaly was discovered with Newton’s system, there was no-one arguing that we should change nature’s laws to make it look more like our model. When some grand theory appears to be refuted by experience, there’s usually no shortage of people arguing that we haven’t tried hard enough to implement the theory.

6 Here’s a really insightful discussion of the significance of imperfect competition in the development of economics as a science:

In contrast 21, the Dixit-Stiglitz 22 did not attempt to explain anomalies. … The aim was to make models with imperfect competition tractable.… 23 meant there was a standard way to handle imperfect competition…. 24 there are no general results…. 25 made it possible to have the illusion that economic theorists understood imperfect competition, but this was discovery by assuming we have a can opener. The example was fruitful because, once a lot of people decided to explore the same special case, they could discuss its interesting behavior…. Theory can grow if people agree on core assumptions. This is progress if the assumptions are useful approximations. … I do not think the the development of a new branch of theory is necessarily scientific progress.

I think economic theory was massively improved by the Dixit-Stiglitz example because it made economists outside of industrial organization willing to consider imperfect competition. But… it gave the impression that there were simple elegant results based on assuming imperfect competition similar to those based on assuming perfect competition. There aren’t….

7 Gans, J. G. and Shepherd, G.B., 1994. “How the Mighty are Fallen: Rejected Classic Articles by Leading Economists”, Journal of Economic Perspectives, Volume 8, pp. 165-179 at p. 170.

8 All mention the significance of scale economies in driving trade patterns together with plenty of empirical data and theorising in both plain English and ad hoc partial equilibrium models and schematic diagrams. Some I didn’t mention in my speech include:

  • Kravis (1956),
  • Drèze (1958),
  • Posner, (1961),
  • Hufbauer, (1965),
  • Elkan (1965),
  • Kojima (1965),

9  I’d add that informal reasoning, of the kind we’re using in this discussion, is the master discourse and formal reasoning its servant, or, to use the image of a marvellous book, it’s emissary. For those who are interested, I recommend the introduction given by Russ Roberts in his recent, characteristically excellent,  interview with the author of the 2009 classic, The Master and his Emissary on EconTalk.

10 I’ve offered some somewhat tangential thoughts on this here and here.

11 Brander, J. A. and Spencer, B., 1985. “Export Subsidies and Market Share Rivalry”, Journal of International Economics, 18, pp. 83-100.

12 Krugman, P., 1993. “The Narrow and Broad Arguments for Free Trade”, American Economic Association, Papers and Proceedings, American Economic Review, 83, pp. 362-371.

13 Eaton, J. and Grossman, G. M., 1986. “Optimal Trade and Industrial Policy under Oligopoly”, Quarterly Journal of Economics, 101, pp. 383-406.

14 The two papers used ‘Cournot’ and ‘Betrand’ models of duopoly respectively. A Cournot duopolist maximises its profits assuming its competitor’s output as given whilst the Betrand competitor takes their competitor’s prices as given.

  1. is[]
  2. 1[][]
  3. 2[]
  4. 3[][]
  5. 4[][]
  6. 5[][]
  7. 6[][]
  8. 7[][]
  9. 8[][]
  10. 9[][]
  11. 10[][]
  12. 11[][]
  13. 12[][]
  14. 13[][]
  15. if lt IE 9[]
  16. endif[]
  17. 2[]
  18. Y[]
  19. I[]
  20. You[]
  21. to Einstein’s explanation of an empirical anomaly in Newtons system[]
  22. model of imperfect competition[]
  23. The (unrealistic) assumptions they made[]
  24. But[]
  25. The Dixit-Stiglitz model[]
This entry was posted in Economics and public policy, History. Bookmark the permalink.
Notify of

Newest Most Voted
Inline Feedbacks
View all comments
Mike Pepperday
Mike Pepperday
5 years ago

Hi Nicholas.
The named theories and theorists are next to meaningless to me but you confirm an opinion I have held for several years: economics is struggling to progress.

Apparently, the introduction of theorising “imperfect competition” didn’t achieve an improvement in theory. I gather, though, that the notion of “perfect competition” was fruitful—had yielded effective insights.

The laws of physics come in the form of an equation expressing a relationship between two or more concepts. The concepts are “idealised” theoretical extremes because reality is messy (there is no instance of just two bodies). We should expect this in social science as well: a theory is an expression of an idealised relationship between two or more idealised concepts.

In physics the relationship is of intensity or strength (not statistical frequency) and (though no one seems to talk about it) this is only possible because the concepts are measurable. Mass, time, distance, etc, are measured in agreed units of kilograms, seconds, metres, etc. In social science there are no agreed units of measure so it would follow that theorising in social science has to be in terms of idealised concepts—such as perfect, not imperfect, competition. Apparently, imperfect competition doesn’t have agreed theoretical meaning whereas perfect competition does.

If that is indicative, then social phenomena can only be theorised as all-or-nothing, as present or absent. I understand homo economicus has proved theoretically useful—but 75 percent homo economicus would be without meaning. “Market clearing,” “perfect information,” ditto. Perhaps this means all concepts except money are theoretically useful (or tractable) only in all-or-nothing terms. Are “supply” and “demand” exceptions? I don’t know but it seems to me that recognition of the necessity of theoretical all-or-nothing for social concepts would be a move in the right direction. Crude maybe, but all-or-nothing is a measure of strength or intensity, and it might be clear enough for those in the game to have a sufficiently similar understanding to construct theory which reveals new insights.

So economics has had theoretical success based on perfect competition as a motivator of behaviour. It stalled among the messiness of real human behaviour and trying to refine the concept of competition doesn’t work. What to do? Introduce other extreme concepts.

Economics assumes rationality and individual utility. To get closer to reality, it would need to assume rationality and sociality. I suggest two further motivators to push toward sociality: cooperation and coercion. (I don’t know that I would put the adjective “perfect” in front of them—perhaps just recognise them as extreme absolutes of presence or absence.) As it stands, economics assumes everyone is alike: competing to optimise individual utility. This would change if cooperating and coercing (and competing) were optimising sociality.

Competition, cooperation and coercion do not necessarily imply sociality (they can apply to non-social animals and even plants) but they are components of sociality. And, most important, they are independent of each other. You can be competitive without any cooperation or coercion. You can be cooperative without competing or coercing and you can coerce without competition or cooperation. So the three motivations form three mutually orthogonal dimensions.

Given that they are orthogonal, economics’ assumption solely of competition makes it blind to two obviously significant dimensions of social reality. Are there others? I can’t prove there are no further motivations but I cannot imagine what would be orthogonal to all three of these. Besides, they seem quite enough to go on with.

The emphasis on competition is not misplaced for cooperation and coercion serve competition. Darwin won’t have it otherwise: we cooperate or coerce in order to better compete. So biologically competition is prior. It is possible to be competitive without knowing, or even knowing of, any of your competitors and that probably applies to 99.9% of species. It can arise even among humans (e.g., a job application). So in principle competition implies zero sociality. The other two dimensions are different: you cannot be cooperative or coercive without some social contact and among social animals competition is mitigated and mediated by cooperation and coercion.

These three motivations become social and human if we add an assumption that people have views of them. That is, people have preferences—totally for or against, nothing in between—about these three concepts. Other social animals are hard-wired for their attitudes but we have culture and for us every individual (who is social) must either accept or reject each of the three motivations.

Three binary options makes eight possible kinds of theoretical person. It’s a big leap from assuming there is only one kind of person but it seems to me they have to be dealt with since views of cooperation and coercion clearly exist as motivators and, being independent of each other and of competition, are not being theoretically coped with at present.

Moreover, though competition is biologically prior and, as animals, people are subject to biological exigencies, they don’t necessarily perceive it that way. Some people might see cooperation or coercion as prior. In other words, cooperation or coercion might be socially more important than competition—it’s a matter of perception.

So I am not only proposing two further motivators but am locating all three in the mind—which is ultimately what they depend on. Economic theory would have it (I think) that competition is something that is “out there” in “the economy” and economists and regulators have their ways of measuring it. Given a theory, perhaps they could find ways to measure (and regulate) cooperation and coercion.

As an aside, in game theory experiments the rewards always assume the maximising of individual utility. No experiments (that I ever heard of) reward with increased cooperation or greater coercion. Huge bias.

The introduction of “imperfect competition” seems to have been an attempt to adjust the theory to try to make it better fit reality. Newton’s formulae, though, are not adjusted. Instead there are new theories which better fit the reality. The economic Einstein must somehow introduce cooperation and coercion.

From the way I have cast it, present economic theory should be suitable for the ecology of trees in a forest so it is obviously overdrawn. Trees don’t have money which allows savings and investment and guarantees of the future. And money can only exist with cooperation and coercion so these things are being snuck in somehow. Yet theory has stalled. Could progress be made if they were explicitly theorised?

Alex Coram
Alex Coram
5 years ago

Hi Nicholas,
The general drift of your assessment is fine with me, although on the whole I would avoid theorizing about what can and cannot be done with mathematics. Two points. 1. I find any attempt to draw analogies between physics and economics troubling. The objects they are dealing with are completely different. In economic models the simplifications don’t apply outside the model. In Newton’s model the dynamics run into the implications of the Poincare-Bendixon theorem with more than two bodies and the problem isn’t with the simplifications. 2. You say the essence of your argument is formal theory should develop in close dialogue with discursive reasoning. Well maybe so and maybe not. Sometimes discursive reasoning is misleading and just messing around with the formal structure of a problem leads to insights that only make sense after torturing the equations. Other times there are no insights that could not be arrived at with discursive reasoning.

Mike Pepperday
Mike Pepperday
5 years ago

It stays on my mind so I’ll have another go. Competition isn’t a theory. It’s a concept, like mass, or velocity or phlogiston. It’s a concept understood by theoreticians who relate it to other concepts so forming a theory. From that they might deduce something which might be useful.

From your reference 5: “Theory can grow if people agree on core assumptions. This is progress if the assumptions are useful approximations.”

First sentence correct; second sentence not. This misunderstanding pervades economics, political science, and no doubt other social sciences. Theory is not, cannot, be built on approximations. It is fundamentally wrong: you do not want approximations to reality when building a theory. Theory is built on concepts which are extremes. Perfect competition was fruitful; imperfect competition was not. If approximations were useful you’d have homo-mostly-economicus, market-partly-or-sometimes-clearing. Those would be approximations. The all-or-nothing theoretical concepts are never realised in reality and they are the ones that work for theory.

Hello Alex! Long time no see. The social sciences, permeated with physics envy, plead to be cut some slack because they are somehow different. But is that reason to actually deviate from the method of the hard sciences? For instance attempting to theorise with concepts that are not idealised?

Galileo theorised gravity with a perfect sphere rolling friction-free on an perfect plane. That is not an approximation to landslides yet his theory is needed to understand landslides. Galileo said the period of a pendulum was independent of its amplitude. His patron, Guidubaldo del Monte, experimented and said it was wrong and when Galileo responded that his pendulum had a pivot with no friction, a string which was weightless and a weight that had no size, del Monte made fun of him. But whose name do we remember? Newton calculated the motion of two bodies. Never occurs. Yet it stands; that calculation is required. (Further bodies are perturbations of it.) Newton’s first law says a body moves at a constant velocity in a straight line forever. There’s not a single example in the whole universe yet this is essential to the calculation of everything from raindrops to galaxies.

Economics uses “full information;” has anyone tried to create theory using “partial information”? I bet it failed to produce anything. “Imperfect competition” is not even phlogiston for phlogiston would have been automatically idealised like every concept in science theory. Imperfect competition can never be idealised.

Yes, theory depends on people agreeing on core assumptions. That would be why idealisation is needed. The implication is that because there are no measurement units of social science concepts, the only measurement that provides sufficient intersubjective clarity to construct theory is total presence or absence.

So what is the oxygen that is to replace the less-than-phlogistonic imperfect competition? I made some suggestions. In the other post you linked to, Nicholas, you make the case that selfishness is of more weight than altruism. It is a case economists have made for centuries. I think they are the only people who express it in such stark terms. It is not cutting through; most people don’t believe it, and though it did advance in recent decades (probably largely through the efforts of Rand and Friedman), it seems to have generated a backlash. I mean, a few years ago a presidential candidate who openly espoused socialism was unthinkable.

At any rate, despite economists’ best efforts, most people will have it that cooperation (as I would express it) is more powerful than competition. They will, like you, submit those data which promote their case—say: abolition of slavery, bear baiting, and child labour. They would argue that without altruism no child would survive, that no soldier would go to war. And they won’t accept that climate scientists are telling fibs to get grant money, or that teachers and nurses are really in it for the dough. These people today campaign against coal mines, whaling, and the harvesting of forests, behaviour which cannot be explained in terms of selfishness. Whether or not cooperation is more important than competition, it clearly has economic impact.

The stasis in economic theory is now two generations old. I suggest that as long as it doesn’t take cooperation on (in idealised all-or-nothing terms) the stasis will continue. And, of course, the same goes for coercion.

Alex Coram
Alex Coram
5 years ago

Hello Nicholas,
Maybe discursive reason is a master discipline (?) in that if we have a sufficiently large capacity we could just use the definitions and intuit their implications. I find if you try to do this with a lot of questions it doesn’t work very well.

Hello Mike,
Good to hear from you. I know the physics envy thing turns up all the time. I suppose it is a statement about mental structures and I try to avoid these. From an n of one (me) mathematics isn’t driven by physics envy or anything else to do with physics. In fact the whole formal structure of social science, insofar as we have any at all, is nothing like that in physics. I take models as aids to thinking – they are not meant to represent any reality beyond their toy world.
Cooperation? Well it depends on what questions you are asking (what aids to thinking you are constructing that is). On the formal side we have a lot of stuff in cooperative game theory. This is what von Neumann was interested in.

Mike Pepperday
Mike Pepperday
5 years ago

Now you mention it—it hadn’t occurred to me before—game theory depends on all-or nothing: defect or cooperate. As you say, it’s a toy world; a real person is “very cooperative” or “quite uncooperative.” And game theory has led to insights. One is the idea of the theoretical evolutionarily stable state. This is some proportion of defectors to suckers who, note, are not half-hearted. If there are “grudgers” these, too, either fully defect or fully cooperate in some stable ratio.

Idealisation has been known a long time. Galileo said with regard to the law of parabolic motion of projectiles: “I grant that these conclusions proved in the abstract will be different when applied in the concrete and will be fallacious to this extent, that neither will the horizontal motion be uniform nor the natural acceleration be in the ratio assumed, nor the path of the projectile a parabola.” Newton: “In philosophical disquisitions, we ought to abstract from our senses, and consider things themselves, distinct from what are the only sensible measures of them.”

Mach and Kaufmann thought idealisation to be universal in science theorising and Hempel and Schütz thought it important. It seems less discussed in recent times.

Montesquieu and Weber thought idealisation necessary for social analysis but their ideal-types were ad hoc. Idealisations aren’t chosen by the theorist; science is after the truths of nature so nature has to decide them—even though they do not actually occur in nature.

Alone of the social sciences, economics developed a body of theory and I am suggesting that one reason for its success is its (unwitting) idealisation of concepts.

If economic theory has not progressed since WW2, the reason may be that the idealisations that had been fruitful were exhausted. This theoretical brick wall and the advent of computers led economics (and much of the rest of social science) into data processing. That has its uses but they are administrative. Big data didn’t generate theory. This is not that surprising: you could not measure everything you thought important about landslides, put it all in a computer, and get the theory of gravity. Science theory doesn’t come from reality; it comes from a human mind, evidently one which ignores data, one which posits unrealistic idealised relationships between unrealistic idealised concepts.

I offered “perfect cooperation” and “perfect coercion” as possible new concepts and gave my reasons. The economic brick wall is now overgrown and covered in moss and a lot of very clever people have not made a dent in it. Economic theory surely has to find some new idealised concepts if it is to advance.

Alex Coram
Alex Coram
5 years ago

Hello Mike,
Glad you are still interested in this stuff.
Game theory has a lot of problems and I have mostly moved on to other mathematical questions. The problems are not what you think, however, and your statement that strategies are all or nothing is completely wrong. This is the case in simple 2×2 discrete games but not generally. Some games have an arbitrarily large set of discrete strategies, others have continuous strategies, some deal with coalition payoffs, differential games have continuous strategies that change at each instant in time and so on. For the last see for example ‘the homicidal chauffeur.’ One of my favourite pursuit evader games.

Alex Coram
Alex Coram
5 years ago

Glad you are still interested in this stuff.
Game theory has a lot of problems and I have mostly moved on to other mathematical questions. The problems are not what you think, however, and your statement that strategies are all or nothing is completely wrong. This is the case in simple 2×2 discrete games but not generally. Some games have an arbitrarily large set of discrete strategies, others have continuous strategies, some deal with coalition payoffs, differential games have continuous strategies that change at each instant in time and so on. For the last see for example ‘the homicidal chauffeur.’ One of my favourite pursuit evader games.

Mike Pepperday
Mike Pepperday
5 years ago

Thanks Alex. Two points.

1. The homicidal driver is not social science. It is maths. (Compare with the trolley problem or the ultimatum game or the prisoners’ dilemma which are moral matters.)

2. If there is a social science theory that uses concepts (concepts, rather than strategies) which are not all-or-nothing and which has delivered insights, I do most sincerely want to know of it. My claim is wide open to falsification.

By “insight” I mean something comparable with the evolutionarily stable state concept. That is not a high bar: I think the ESS (which is now at least a generation old and is still pretty esoteric) is just theory; I don’t know that it is testable or led to any practical applications but it did reveal possibilities. Another example from social science is the saw ‘democracies never war against each other’ which is testable – and is all-or-nothing: democracy or not, war or not. No degrees, no “warlike.”

Anything. Anything where social science succeeds where science apparently cannot: using concepts that are not idealised. I’d like to know of it.

Mike Pepperday
Mike Pepperday
5 years ago

I am trying to figure out what’s going on. Nicholas disagrees with me but admires my confidence which is really saying I am overconfident—or dogmatic or something. And Alex says I am completely wrong. Completely. Now that’s confidence.

The only original assertion in my disquisitions above is that effective social theory requires all-or-nothing, extreme concepts and that moderate, realistic concepts cannot generate useful social theory. I don’t actually have any reason not to insert “always” and “never” at the appropriate points in the last sentence—other than lack of confidence.

The “completely” is factually false since I gave illustrations which indicate my claim is at least sometimes correct. So why so emphatic? Would allowing partial correctness leave a chink in the armour? What armour? Career-long assumptions, I suppose.

My assertion is very simple. Is it that it is hard to accept because if it had merit, you’d already know of it? This is reason for suspicion but it is not a refutation. It would just reinforce the complaint that the older generation has to die in order for a new idea to progress.

Nicholas, you are thinking all the time about it and you recognise there’s a fundamental problem with economic theory. My diagnosis would have significant implications. If it is completely silly, why not show it and thus rule it out?

Alex, why not omit the “completely” and declare: “You are wrong, for the following reason(s)”? In principle you only need to produce a single falsifying instance.

Please sort it out, if not for your and my sakes, for all the innocent lurkers reading Troppo who could be taken in by my silver tongue.

Alex Coram
Alex Coram
5 years ago

Hi Mike,

The discussion goes
Pepperday: ‘game theory depends on all-or nothing: defect or cooperate. ‘
Coram: ‘your statement that strategies are all or nothing is completely wrong’
I said why with examples of strategies that are not all or nothing.
This is all I said. I concede the completely is redundant and will just go with ‘you are wrong’ in the same sense as if you said pi = 7 is wrong.
It is nothing to do with career long assumptions.
By way of extension pursuit evasion games are important for military strategy, evolutionary stable stated depends on a probabilistic strategies, etc.

In continuation you introduced qualifiers about insights and reference to concepts and strategies which are unclear.

It is possible that these distinctions and related discussions about meta-theory have value but as I said at the get go I avoid theorizing about what can and cannot or should and should not be done. I am sorry if this makes me pretty useless as a correspondent on philosophical matters.

5 years ago

Mike, I’m not sure where the boundary between science and social-science is, although based on previous conversations you’ve had, I assume you mean “hard science” for science.

Thus in terms of :”By “insight” I mean something comparable with the evolutionarily stable state concept” how about the the Hodkin-Huxley equations, which form the bases of large amounts of computational neuroscience ( This has given huge insight into things from what they started trying to explain to how electroencepholography rhythms your brain produces emerge from cell populations.

5 years ago
Reply to  conrad

For historical models that are simpler to undestand, you could have the Papez circuit

paul frijters
paul frijters
5 years ago

Me and Mike agree 80%. I thus agree theory is mainly about relations between concepts, and that key concepts should abstract ruthlessly.

I have no hope for the remaining 20%, because I do want my theories to be guided by stylised data, purpose, and historical narrative, whilst Mike seems not to because it is not absolute enough. Fortunately, the issue is moot.

Mike Pepperday
Mike Pepperday
5 years ago

Thanks Nicholas. You say: “You think social sciences are essentially like natural sciences and we are seeking to pursue knowledge using the ideas of abstraction that worked for Newton in gravity.”

To clarify: I don’t quite think that. I see the hard sciences as giving us suggestions, tips. Their success qualifies them for this. And if science always (or just mostly) obeys a stricture, we would need to have good reason before we assume social science need not obey it.

The big difference between social and natural is that social science facts are not objective facts so there is a severe problem of what is real. Is temperature a real thing? Yes, because we know (for example) a relationship between it and the expansion of mercury therefore we can make an instrument to detect it, therefore temperature is a real thing.

In science, if it relates, it’s real. If you can’t measure it (at least in principle) it can’t exist for science. So phlogiston and the aether fail. We cannot detect social phenomena with an objective instrument. Two scientists in two different times and places can talk about temperature and be talking about the same thing. A Likert scale doesn’t provide that intersubjectivity.

Surely there will never exist an instrument which you can point at a building and have it tell you whether it is an asylum or a university. Or at some objects and have it tell you they are money, or at two people talking and have it tell you they are playing chess. The asylum, the money, and the chess are facts but they are social facts. They not objectively measurable because we have no relationship between the manifestation of the phenomenon and some indicator analogous to mercury. Social facts ultimately exist only in minds, by agreement.

Ronald Searle spun a whole book out of that (The construction of social reality) but he missed idealisation. He thought science dealt with real things but it doesn’t. As far as I can tell, science theory absolutely always deals in idealisations: idealised relationships between idealised concepts. Idealisation even extends to the practice. The idealised theory predicts some temperature and then impurities, bad hair, whatever, mean the reality deviates. You stick the thermometer in and read it: 75.7 degrees. You write it down: 75.7—and you have turned reality into mathematical perfection. The act of recording idealises. The textbook tells you some temperature and you make a judgement as to whether your reading conforms. Intersubjectivity reigns because you have units of measure.

That isn’t possible with social phenomena. An instrument to measure levels of emotions and moral values would be an instrument that measures nerve activity but I would say the present state of that art, Conrad, is much more biology than social science. Measuring the kilos of competition or the litres of cooperation is nowhere in sight.

And so I suggested there is a chance of intersubjectivity by abstracting social phenomena to the extremes of present or absent. I offered quite a bit of supporting evidence. That evidence included some of the concepts of economics, among them that “imperfect competition” doesn’t work whereas “perfect competition” does. It is this idealisation, I submit, that has allowed economic theory to blossom. The extremes of presence or absence are sufficiently intersubjective for those in the game to be on the same page and useful theory to be constructed.

I am pretty confident my all-or-nothing thesis generally holds but I am particularly interested in economics because it is the only social science to have developed a body of theory (though you might make some argument for linguistics and law). The success of economics is also due to the concept, and existence, of money and I don’t have anything to say about that. Somehow the extreme absolutes of perfect competition, market clearing, homo economicus, etc are adapted to give rise to the continuum which is price. I have no idea how that works.

80% agreement between Paul and me is dead suspicious and perhaps I can cut it back a bit. “…key concepts should abstract ruthlessly.” This implies that abstractions are a matter of choice. If abstractions mean idealisations (which I think here they do) let us note that they are not freely chosen in science. Some philosophers (and economists) think idealisations are chosen to simplify. No. Nature chooses. If there are other technical civilisations in our universe they will have exactly the same formulae for Newtonian mechanics as we do. The idealisations do not actually exist and yet Mother Nature dictates what they are. Question: Would those technical civilisations have found the same relationships as Smith and Ricardo or that “wave of formalism in economics from around 1870 to World War II.”?

Of course, reduction to all-or-nothing extremes is not evidence a concept exists. Only testable relationships provide that. But I have tested that Troppodillian attention span enough.

John R Walker
John R Walker
5 years ago
Reply to  Nicholas Gruen

dont know why, but this conversation for some reason really reminds me of a core theme of Gödel, Escher, Bach Chapter XX :
” Can Machines Possess Originality?”.
Hofstadter’s answer to that question is yes.

Mike Pepperday
Mike Pepperday
5 years ago
Reply to  John R Walker

John, your copy and mine disagree: my Chapter XX is titled “Strange Loops, Or Tangled Hierarchies.”

Chapter 22 of “The Mind’s Eye” (which I found more interesting than “Godel…”) is by John R Searle and called “Minds, Brains, and Programs.” That it? It’s the thought experiment about chinese translation. Not much relevant to this thread though.

I mention Searle above but wrongly gave him the Christian name Ronald.

John R Walker
John R Walker
5 years ago
Reply to  Mike Pepperday

Hi Mike
Sorry I should have said,
“Can Machines Posses Originality”
is the first subheading of Chapter XX of GBE.


[…] Gettler interviewed me recently on my exchange with Krugman. As you can imagine, it’s a difficult thing to explain in an interview, but I took that as a […]

Mike Pepperday
Mike Pepperday
5 years ago

Thank you for going to such trouble, Nicholas. It’s true I don’t understand the details. Still, if these trade hypotheses which rival each other don’t lead anywhere then theory is not progressing—which is where I began: economic theory has run out of puff.

Looking back to an earlier post: “…while a few big moves can be made by pursuing Adam Smith’s idea of exploring the tendencies that competition brings about, beyond that the game is mostly trying to find patterns…” And similarly: “Most of the ‘theory’ that can be useful is done and most benefits will come from other activity,” and you advocate commonsense.

So your position is clear: competition made effective theory but it is now exhausted. If, as you say, no more theory is possible then indeed all that is left is to be discursive and use commonsense. That is how most of the human race functions and how all of it did until Galileo et al came along.

It’s a worry for commonsense said the earth was flat and it was theory that sorted it. I think further useful economic theory may be possible—if we obey the idealisation rules of science theorising. The biggest (and unrecognised) stricture is to use all-or-nothing concepts as was so successful with competition and which, I claim, is the only effective way to theorise when there are no agreed units of measure. If those rival trade theories were not all-or-nothing then, like attempts to theorise with imperfect competition, they were doomed from the outset.

I can’t know every social science so thank you for trying to educate me. Thanks to Alex as well. I was a bit unfair on him. I apologise. I have been on the lookout for evidence which might falsify my claim for several years. We went down a few tracks on this thread but I still know of no falsifying instances. Perhaps, though, I have learnt that persuading people of my claim is not easy.

Mike Pepperday
Mike Pepperday
5 years ago

All-or-nothing is not well defined and imperfect competition is. Sure. But it is no good trying to depend on definitions. It won’t work. We had a discussion on this a year or three ago. The only place meaning can decisively come from a human definition is with human things such as contracts, laws, and measurement units. Science tries to figure out what mother nature is doing, so it defeats the purpose to try to impose human opinions on her.

By “all-or-nothing” I just mean something that’s there or it’s not. Are there theories that don’t mention competition? In those, it’s not there. There are theories with homo economicus and market clearing and presumably there are theories that don’t use these concepts. That’s all there is to it: a concept is present or it is absent. It is not half present. When you read the thermometer as 75.7 everyone understands it the same. There is no homo-fifty-percent-economicus (which would be more like the reality) for the reason that there’d be insufficient intersubjectivity; if people can’t talk and understand each other, theory can’t progress.

If “imperfect competition” is “some competition” the problem is with “some.” My all-or-nothing injunction forbids it. Those hundreds of thousands of economists aren’t aware of the all-or-nothing rule. They obey it unwittingly (or used to until they wandered off into “imperfect competition” etc). Like grammar: everyone obeys without knowing the theory. Every person, plant and microbe obeys Darwinian natural selection whether they know of it or not. “Democracies never war against each other” operated long before anyone noted it. People act in accord with a theory but aren’t aware of it.

The need is to recognise: (1) the necessity for ideal-types; (2) that these must be discovered, not invented; (3) that in social science the only ideal types available are presence and absence because there are no measurement units.

And prior to those, to recognise that a scientific theory expresses a relationship between two or more concepts. Is that a statement of the obvious? I think it ought to be but I don’t see anyone saying it, let alone that it is obvious. And I do see Chalmers (What is this thing called science?) discussing “statements” (not “relationships”) about white swans and immortal Socrates which do not relate anything to anything.

AFAIK all science theory deals with, and only with, ideal types. Economists are not aware that effective theorising uses idealised concepts. They think their concepts are simplifications. I can quote Krugman on that. Philosophy of science is vague about idealisation, suggesting that ideal-types simplify and can be chosen by the scientist. There are a couple of online philosophy encyclopaedias and I once looked into it. Hubris. Nature chooses them and it is up to the theorist to discover them.

Weber is famous for pushing ideal-types in social science. I am not the first to complain that he invented his ad hoc yet these complaints haven’t stopped philosophers from thinking they can be invented. As I noted above, John R Searle apparently doesn’t even know of idealisation. (Presumably he would but doesn’t see the import.) He is one of the world’s famous philosophers and that book and its punning title is well known.

One problem with thinking you can invent concepts to “simplify” is that leads to supposing the simplifier might be the typical or the average. But ideal types are extremes. Weber has a point in saying:
“The sharper and more unambiguous the ideal-types are made, that is, the more foreign to the world in this sense they are, the better they perform their terminological, classificatory and heuristic service.” (emphasis original)
But in the end, they are what they are. Enquire of mother nature; she “defines” not with words but with ideal-types (Platonic forms if you wish). She conceals them under layers of colourful, messy, distracting reality but if you approach her right she reveals them. Pick two (or more) idealised concepts, propose a relationship, and deduce the consequences… (Is that what you mean by “formal modelling”?)

As a unit of measure all-or-nothing is crude: if you ask how much competition there is, the answer is either “present” or “absent.” This “measurement unit” is all that is available to social science and the record is that it does work. The other disciplines haven’t woken up to it—except as I mentioned, “democracies never war against each other” which seems to be the only definition-free, falsifiable theory they have.

To falsify: point to a useful social science theory that uses partial concepts. I pointed out 2×2 (all-or-nothing) game theory produces useful insights. Alex says bigger matrices are partial. That’ll do: What do they yield?

Mike Pepperday
Mike Pepperday
4 years ago

I agree with Marshall’s 1925 warning of statistics doing more harm than good. He wasn’t alone and I have a small collection of comparable early warnings. Here they are.
Kurt Lewin railed against statistics in 1931:
“[Counting frequencies] reduces one to a treatment of these problems in terms of mere averages, as exemplified by tests and questionnaires. Any one to whom these methods appear inadequate usually encounters a weary scepticism or else a maudlin appreciation of individuality and the doctrine that this field, from which the recurrence of similar cases in sufficient numbers is excluded, is inaccessible to scientific comprehension and requires instead sympathetic intuition.”
[K. Lewin, “The conflict between Aristotelian and Galilean modes of thought in contemporary psychology,” Journal of General Psychology, vol. 5, pp. 141-177, 1931, page 155.]
Psychologist JG Taylor said in 1958:
“A law of nature is not a catalogue of the parts that every system must have, or a classification of the parts according to the frequency of their occurrence, but an explanation of how the parts hang together in the system.”
[J. G. Taylor, “Experimental design: a cloak for intellectual sterility,” British Journal of Psychology, vol. 49, pp. 106-116, 1958, page 110.]
Exactly: how the parts hang together is real knowledge. Goethe saw the problem 200 years ago; he has Mephistopheles saying:
Who something living would know and describe
Seeks first the spirit from it to drive.
He then has the parts all in his hand;
Missing—alas!—just the spiritual bond.
Taylor drew an analogy of trying to work out what made a car’s speedo needle move by looking to correlations of the various input pedals and levers. You would wind up deciding the turn indicators were correlated with the needle dropping but the device that would most “account for” (the usual term in statistics papers) the speedo would probably be the steering wheel—the less it moved the higher the needle.
Millions of not particularly bright people can give a correct explanation of the speedo because they know the “spirit” connecting the parts. Presumably social and economic systems are more complex than motor cars. Why do we expect factor analysis of inputs and outputs to yield understanding? The two strongest theories in psychology would be learning theory and behaviour theory which were not developed by statistical analysis. Nor was utility in economics.
Neither the Linnean taxonomic system nor the periodic table of the elements were developed with statistics. An anachronism? No: four hundred years ago Francis Bacon started counting the associations of warmth and dryness. It was never going to get him anywhere. Yet today this is how much social science proceeds. This is the status of the “Big 5” personality “theory” which would have thousands of papers supporting it along with the inputs and outputs of maybe a million tested subjects. Rubbish.
In 1967 Hayek took up the theme using computers as an analogy:
“Statistics might assist us where we had information about many complex structures of the same kind… but it presupposes that we have an independent criterion for identifying structures of the kind in question…
“How little statistics can contribute, however, even in such cases, to the explanation of complex phenomena is clearly seen if we imagine that computers were natural objects which we found in sufficiently large numbers and whose behaviour we wanted to predict. It is clear that we should never succeed in this unless we possessed the mathematical knowledge built into the computers, that is, unless we knew the theory determining their structure. No amount of statistical information on the correlation between input and output would get us any nearer our aim. Yet the efforts which are currently made on a large scale with regard to the much more complex structures which we call organisms are of the same kind. The belief that it must be possible in this manner to discover by observation regularities in the relations between input and output without the possession of an appropriate theory in this case appears even more futile and naive than it would be in the case of the computers.”
[F. A. Hayek, “The theory of complex phenomena,” in Philosophy of social science, M. Martin and L. C. McIntyre, Eds., ed Cambridge, MA: MIT Press, 1994 [1967] page 59.]
Hayek thought this “futile and naive” process was “large scale” in 1967! Millions of papers later the statistics train thunders on into the darkness.
One prominent source of statistics papers is economics. For administration purposes statistics are vital and they may also play a role in the testing of theories. But without theory—which is the situation elsewhere in social science—statistics and correlations lead nowhere.

paul frijters
paul frijters
4 years ago
Reply to  Mike Pepperday

we have had this discussion before Mike, but you are too absolutist in dismissing the “count and discover” aspect of social science where data precedes theory, as in discovering cosmic background radiation and the bacteria where theories of what was being observed followed long after simply using the measurement tools and describing plus classifying.

Interestingly, you say “Neither the Linnean taxonomic system nor the periodic table of the elements were developed with statistics. ”

eh, really? I am no expert on how they came to be, but this sounds very doubtful. On the periodic table, one related website ( tells me
“In 1869 Russian chemist Dimitri Mendeleev started the development of the periodic table, arranging chemical elements by atomic mass. He predicted the discovery of other elements, and left spaces open in his periodic table for them.”

how is ‘arranging chemical elements by atomic mass’ anything else than ‘developed with statistics’? The atomic mass of an element is a statistic.

Science involved thinking up theories as to how entities (often completely imaginary) relate, but it is not exclusively about relationship between ghosts. Sometimes we are indeed administrators peering at data we dont understand, slowly seeing patterns for which we then conjure up the ghosts that we assume gave rise to them. As Sherlock would say, one needs data to make theories.

you also say “without theory—which is the situation elsewhere in social science—statistics and correlations lead nowhere.”

eh, again, too absolutist. If you are just interested in prediction, which we often are, then correlations and statistics are very useful. If I want to know if winter is coming, it is not anti-scientific to see which way the birds are flying, even if I lack any theory as to why their movements precede winter. Simply knowing they do is knowledge that is of use. Its not the end of the endeavour as it makes for a poor theory, but of course which theory has ever proven to be the final word that wasnt eventually overturned by …. data ?

Alex Coram
Alex Coram
4 years ago

Not sure about these meta-theory debates. Sometimes statistics is useful in giving insights and testing an idea. Sometimes it isn’t.

Mike Pepperday
Mike Pepperday
4 years ago

Hi Paul. If I’m absolutist, so are the people I quoted. They were there before me. The evidence is now in and they were right: statistics as practised in social science has led nowhere. Millions of statistics-based soc sci papers with no useful result.
“If you are just interested in prediction, which we often are, then correlations and statistics are very useful.” I didn’t say statistics weren’t useful. On the contrary, I said they were vital: statistics are vital to administration. And the very purpose of collecting them is to predict.
But I am not “just interested in prediction”. Nor was Hayek, Lewin or Taylor. I expect they were well aware of the usefulness of statistics but they, like me, were interested in understanding, which statistical analysis could not deliver. The evidence is now in: it didn’t.
Actually, I think you are incorrect about cosmic background radiation. Wikipedia says it was first theorised in 1948 and discovered in 1964. But it’s beside the point. Presumably many a theory resulted from puzzling over observations but does data precede theory? This is not my view and it’s not the mainstream view which is of the “theory-dependence” of observations. Without theory, at least a hunch, you can’t collect data because you don’t know what to collect. Serendipity may sometimes occur—Roentgen noticed an image on photographic film. Even there, he noticed because he had a prior expectation.
Collecting data without a theory might describe many statistical projects in social science. I call it automated, theory-free induction; someone described it as torturing the data till it confesses.

I am familiar with the Mendeleev story. He labelled cards with the (known) elements and played “patience” with them. A few people had tried this approach of seeking patterns but he was the one to succeed. From the relationships so revealed he made predictions of missing elements. This is standard science: you posit the relationships to make predictions which you then test. I think atomic mass is atomic mass, not “a statistic.” Be that as it may, he certainly didn’t do any statistical analysis. It seems the reason he succeeded was that he had an exhaustive familiarity with the properties of the elements so shifted the cards to best match properties down and across.
To infer the coming of winter from the birds is neither pro- or anti-science. Induction allows us to believe in a flat earth and predict the sun will rise tomorrow. Bertrand Russell had an inductivist chicken which predicted it would get fed at the same time every day—until the day it lost its head when it had grown fat. The chicken’s statistical analysis did not provide correct understanding.
I think statistics never provide understanding. You could take the current data of the chemical elements and put them in a statistics program and you will never see the periodic table. (Incidentally, the table is not undisputed. For example, the elements may be distributed around a tetrahedron or a cube.) Or imagine a curve drawn on a monochrome screen—a parabola, say. Input the x,y coordinates of the pixels that are lit to a statistics program but the program will never deliver the formula of that curve. A high school maths student can tell you but statistics will not. Fourier analysis could but not statistics. As another example: measure every conceivable parameter of a million autumn leaves and input the data to a statistics program—it will never produce a theory of aerodynamics—yet the leaves obey aerodynamic theory.
As those examples, and the examples of Taylor’s speedo needle and Hayek’s computers show, relating things by frequency of occurrence does not reveal the real interrelationships (if any). It just can’t. Incidence counts do not have anything to do with any actual relationship; all they can yield are correlations, not relational theory.

The basis for this meta-debate is that social science has failed—except economics which has stalled. My old posts above make a simple point about the all-or-nothing nature of economic concepts. This is a feature that is not recognised. I didn’t succeed in persuading anyone though I don’t see why for it appears obvious to me. A scientific theory expresses an idealised relationship between two or more idealised concepts, and all-or-nothing is the only way to idealise social science concepts because of the lack of measurement units. That is, all-or-nothing is the only available idealised quantity (quantity, not count). I submit that this idealisation is a crucial reason economics came to rule the world.
My further point was that economic theorising fizzled out about two generations ago. Since “imperfect competition” had gone nowhere (a confirmation of my all-or-nothing thesis), a possible approach might be to stick with all-or-nothing competition and to add to it all-or-nothing coercion and cooperation.
The other social sciences have produced no body of theory. That may be one reason people so enthusiastically embraced statistics. But the statistical approach has also failed. I suggest interrelating all-or-nothing idealised concepts. This did, after all, work well for economics.
* * *
“And by ‘statistics’ I was reading it to apply to ‘formal models’.”
You would have to give me an example, Nicholas. As far as I can see, when you count and correlate there is no room anywhere for models.

paul frijters
paul frijters
4 years ago
Reply to  Mike Pepperday

“he had an exhaustive familiarity with the properties of the elements so shifted the cards to best match properties down and across”

yes, we nowadays call that ‘torturing the data till it confesses’ :-)

I do agree there is a role for idealised theory and that economics is currently not innovating as much as it should, essentially because the field has become too big and has atrophied in its core tenets.

Yet, I do think you have the role of statictics and imperfect competition wrong in the history of economics. One of its most successful products, the market cross, is an imperfect competition story, and one of its most successful heuristics is the to-do list in case of various imperfections in markets. Those products and heuristics go way back. And a large role in the success of economics has been the advent of GDP – bean counting, precisely because it appealed to the idea that we could now steer the economy as we steer the building of a bridge. GDP a very leaky bucket of statistics, with as underlying principles that one is trying to measure how much a country can (within a few months) orient towards a war effort, and to studiously avoid counting the things women did. And to make your point: its introduction has probably been very bad for our understanding of the economy and for our policy advice. But I repeat the point: (economic) science is not as simple as saying that only idealised theory is understanding and that playing with statistics is a dead end.

paul frijters
paul frijters
4 years ago
Reply to  Nicholas Gruen

“The master discourse is discursive reasoning”

a big claim. Master discourse, hej?

paul frijters
paul frijters
4 years ago
Reply to  Nicholas Gruen

an example of saying “if you cannot show me B, A must be true”. Like science is not merely about idealised theory, so too is discourse not restricted or necessarily ‘Mastered’ by discursive reasoning.

For instance, if a picture says more than a thousand words, aren’t pictures high in the discourse stakes?

paul frijters
paul frijters
4 years ago
Reply to  Nicholas Gruen

comment image

You use terms like discursive reasoning as if its clear what that means, and also that it is Master amongst some set. Not clear to me at all. A dance indeed is much more convincing.

I also reject the notion that this is what you or I do. If I had to describe how we discourse about specific problems, its often more trying to find some previously known pattern/story that fits, where the ‘discourse’ is about pointing out the bits in the pattern that fit and other bits that fit less.

Is that discursive reasoning though? The definition I keep finding is

Discursive reasoning is thinking a problem through logically step by step from one premise to another in an attempt to arrive at an acceptable conclusion or explanation, as opposed to intuitive knowledge.

Logically. Premise. Step by step. Hmmm. Is loose pattern recognition ‘logically step by step’? Not really. Acceptable conclusion/explanation yes, but thats the point of any discourse and the key thing is that acceptable is a social thing.

So if I follow the common definitions of it, I have to disagree the discursive reasoning is master of much. I find myself wondering if mathematics is discursive reasoning, initially taking it more to be a kind of intuitive reasoning (because its a bunch of ‘if then’ statements where everything of value is the intuition that tells you if the ‘if’ bit is true or not. The step inbetween might be logical, but the if bit is belief). I then find lots of websites telling me there are lots of different philosophies on what mathematics is, and I find no agreement.

Now of course can I redefine all discourse leading to some suggestion of a conclusion as discursive reasoning, in which case it is both the Master and the slave discourse and everything from dance to mathematics and hymn is included.

So I basically just disagree with you in terms of your claim that discursive reasoning is the Master discourse. Perhaps ‘pursuasive argumentation’ or even ’emotional resonance reasoning’ are the discourse with greater effects. Perhaps in general I think story telling and pattern recognition is what we do. But stories are not really discursive reasoning. Is pattern recognition discursive? Not clear to me.

paul frijters
paul frijters
4 years ago
Reply to  Nicholas Gruen

Ni Nick,

ok, that’s a bit clearer. Would it be fair to say you think of discursive reasoning as something like “Trying to come to a common understanding with someone else by presenting ones reasoning openly using shared ‘common’ language, concepts, and simpler stories”? Which means there is some notion of a conversation in there whereby one tries to get the other to join some mental train of thought via shared words, patterns, stories, whatever?

I guess I would agree that that is a more useful and more powerful discourse for common discovery and sharing of ideas than most other forms of communication. Whether that is what happens a lot or is The most powerful way, it really is tough to say. Ideas often get taken up if they are seen to be successful, which can be because the person who adopts them is seen to be successful. The ‘conversation’ is then no more than noticing the success and the association with some idea. No verbs and any overt attempt at communication needed. Also, different groups will use different language. Mathematicians send each other proofs without explaining what their concepts ‘mean’ outside their language. No further meaning needed of wanted in many cases. 2+2=4 needs no further explanation as to what is meant with 2, +, =, or 4.

I guess as an empirical statement, its probably true that written stories remain the dominant way to convey ideas and thinking to others over time and space. Stories often include images, smells, emotions, etc. though they derive meaning outside of the written word and need no words to be described. Pictures and videos within the stories work better.

Synchronising thought processes in large groups is probably done predominantly via sound, smell, and visuals, as when a crowd in a stadium behaves as one. They are not reading the writings of others, but see and hear them. Think also of lovers conveying ideas via touch.

So I do agree that social science, including economics, is mainly story telling and that stories are predominantly communicated to others within the field in common written language. If you want to call that discursive reasoning, fine, but that is not what the dictionary tells me that discursive reasoning is. But then that dictionary definition sounds a bit fantastical.

John R Walker
John R Walker
4 years ago

those below decks would like to know:
What you are talking about and:
What are the practical imports of your ,differences.
Respectfully yours …

paul frijters
paul frijters
4 years ago
Reply to  John R Walker

:-) I ‘think’ that Nicks main point is that mathematical formalism and aggressive posturing do not represent how the main ideas in social science came about, how they are communicated to others, or how anyone becomes convinced of them.

I am not so sure. People are weird animals who take their cue from lots of things, seldom the ‘truth’ or logic, even scientists.

Nick also seems to claim that a particularly lofty form of story telling – where everything goes from logical premises to agreed conclusion – is what we should all be doing and that is the main thing worth doing.

Again, I am not so sure.

So we are basically having a chat about how to chat.

John R Walker
John R Walker
3 years ago
Reply to  Nicholas Gruen

It sounds like masters of the economics training system should get out more often. (And possibly go and have a look at the apophatic tradition in the eastern church).

Important thing about history is as WK Hancock used to say ( over and over) to his students “ history is done with your boots on”.

John R Walker
John R Walker
3 years ago
Reply to  Nicholas Gruen

Was suggesting that reading and traveling widely outside of ones own specialty is what makes the difference .

John R Walker
John R Walker
3 years ago
Reply to  Nicholas Gruen

“not getting the hang of it” true

paul frijters
paul frijters
3 years ago
Reply to  Nicholas Gruen

(nearly) all this I agree with. What you call the framing problem and what I have called the pattern-recognition aspect of social science is entirely ignored as a subject of its own in economics even though that is the core of it: the eventual ‘skill’ that a trained social scientist ends up applying. I agree it should be center stage and that education would be lots faster and thus deeper if we did do that.

Why didnt you say this immediately instead of throwing strange lofty claims around about discursive reasoning? Made you sound like those philosophy add-ons!

I will keep working on improving my Gruen translator.

paul frijters
paul frijters
3 years ago
Reply to  Nicholas Gruen

and I like the word pattern recognition because it immediately makes it clear how primal and how limited what we do is. It immediately conjures notions of arguments about whether something is a circle or a square in the situations where it is not perfectly either one. In contrast, ‘framing’, ‘paradigms’, even ‘ideas’ have this loftier pretentious loading of some absolute truth or ‘best path towards truth’ to them, which I often find, well, deeply untruthful about what is going on.

But let’s leave it there lest our translators overload once more.

Mike Pepperday
Mike Pepperday
3 years ago

If the master is discourse, the mistress is theory and she is being neglected. Discourse had been going for two thousand years when Galileo introduced theory—and changed the world. Let me discourse, then, about theory (not discourse) and about the substantive matter of social and economic relations.
Theory, which expresses idealised relationships between idealised concepts, is not applied in social science except in one area: economics. In consequence economics, alone of the social sciences, has been a success. But economic theory is now exhausted and stagnating. Discourse is churned out in unprecedented volumes—it will never be neglected—and instead of theory we have, all over social science, a raging epidemic of completely useless, post hoc, theory-free, automated induction. Leading thinkers said it would be a waste of time and now, as Marshall put it, statistics are likely doing more harm than good.
My suggestion is to broaden the formal theorising of economics from competition to include coercion and cooperation. These three sorts of social relations are actually widely recognised. Economics routinely recognises three corresponding sectors: market, government and “third sector” (NGOs etc).
Many thinkers have spotted the three. Rick Wicks, then a post-grad economics student at Gothenburg, published a long list of such triples. Here are the ones that I thought the most striking. NOTE The citation footnotes did not carry over when I pasted this post so I have pasted them at the bottom. It is easy to see where they apply.

Kenneth Boulding: “economic, political, and social” spheres; three modes of resource transfer: trade with you, take from you and give to you.
Mackey: “economic, political, and social problems”; “the new political, social, and economic paradigm”; “political, economic, and… cultural control.”
Thomas Friedman: “corporate-led coalitions to create commercial value…; government-led coalitions to create geopolitical value…; activist-led coalitions to create, or preserve, human values.”
Friedland and Alford: “logics of action”: in the marketplace, individual utility and efficiency; in the polity, democracy and justice; and in the family, mutual support.
Irene van Staveren: “three values appear time and again in economic analysis: liberty, justice, and care. Markets tend to express freedom, states to express justice, and unpaid labor to express care among human beings.” She notes (p. 213) that C. E. Ayres asserted a similar set of core human values: “freedom, equality, and security”. Van Staveren distinguishes: forms: exchange, redistribution, and giving; locations: market, state, and the care-economy; virtues: prudence, propriety, and benevolence; development aid: self-reliance, rights, or emergency aid; symbols: Lady Liberty, Joan of Arc, and Mother Teresa.
Waterman: “three freedoms: economic, political, and religious (conscience).”
Hobson: “the democratic triad of liberty, equality, fraternity.”
Bowles: “states, communities, and markets…”
Wright: “governance, moral codes, and markets.”
Steedman: “potatoes, politics, and prayer”. Steedman quotes Philip H. Wicksteed’s 1885 “business, politics, and the pulpit” in his book of sermons Is Christianity Practical?
Minowitz’s book title: Profits, Priests, and Princes: Adam Smith’s Emancipation of Economics from Politics and Religion. Minowitz quotes Trotsky: “God, kings, and capital.”

Wicks gives a number of other examples that employ variable vocabulary and which are less clear though readily defensible.
I myself put together a list of over two dozen categorisation schemes, only one of which overlapped with Wicks’s list. They are not so convenient for me to copy and paste (as I did with the above) but they include famous names in anthropology, sociology, psychology and political science.
There is something there, something objective in human social relations. People interrelate in three principle ways and the three are (in principle) independent. There is an opening here for some lateral-thinking, uncommitted economist to give coercion (compulsion, prescription, duress…) and cooperation (collaboration, collusion, connivance, conspiracy…) the sort of theoretical treatment that competition got. Overall, it is a matter of positing “rational and social” instead of “rational and utility maximising.” Maybe a new theoretical world would open up.

Wicks, Rick. 2008. The three spheres of society – markets, governments, and communities – related to Fiske’s four relational modes: modeled in dynamic balance explaining why social goods are more important than social capital. University of Gothenburg. School of Business, Economics and Law, Department of Economics. Wicks, Rick. 2009. “A Model of Dynamic Balance among the Three Spheres of Society ¬ Markets, Governments, and Communities ¬ Applied to Understanding the Relative Importance of Social Capital and Social Goods.” International Journal of Social Economics 36(6).
Boulding, Kenneth (1978), Ecodynamics, A New Theory of Societal Evolution, Sage Publications, Beverly Hills. Boulding, Kenneth (1985), The World as a Total System, Sage Publications, Beverly Hills. Boulding, Kenneth, Elise Boulding, and Guy M. Burgess (1980), The Social System of the Planet Earth, Addison-Wesley: Reading, MA.
Mackey, Sandra (2002), The Reckoning: Iraq and the Legacy of Saddam Hussein, W.W. Norton: New York, (384, 217).
Friedman, Thomas L. (1999/2000), The Lexus and the Olive Tree, Harper Collins, London, (202).
Friedland, Roger, and Robert Alford (1991), “Bringing Society Back In: Symbols, practices, and institutional contradictions”, in The New Institutionalism in Organizational Analysis, Powell & DiMaggio (eds.), (39), cited in DiMaggio (1994).
van Staveren, Irene (2001), The Values of Economics: An Aristotelian Perspective, Routledge: London, (24).
Waterman, A.M.C. (1986), “Christian Political Economy: Malthus to Margaret Thatcher”, pp. 99-124 in Block & Hexham (eds.), (123).
Hobson, J. A. (1938/1976), Confessions of an Economic Heretic: Autobiography, Harvester Press: Brighton, (52).
Bowles, Samuel (1998), “Endogenous Preferences: The cultural consequences of markets and other economic institutions”, Journal of Economic Literature 36:1 (March):75-111, (105).
Wright, Robert (2000), Nonzero: The logic of human destiny, Pantheon Books, New York, (99).
Steedman, Ian (1994), “Wicksteed: Economist and prophet”, in Brennan & Waterman (eds.) (211).
Minowitz, Peter (1993), Profits, Priests, and Princes: Adam Smith’s Emancipation of Economics from Politics and Religion, Stanford University Press, (240).
Trotsky, Leon (1957), Literature and Revolution, Russell & Russell: New York, (255).

paul frijters
paul frijters
3 years ago
Reply to  Mike Pepperday


I do think the type of theorising you call for is happening – lots of it. None of it sticks though. None of it makes it to the main 1st and 2nd year textbooks which have remained virtually the same since the 50s. And when a challenge is made to the 50s textbook, it calls back theory from even later (like marxism or the Austrians).

Kahneman noted this a few decades back and gave reasons for this similar to what was argued by Hotelling a century or so ago. Economics has become so big and so specialised that the individual branches are nearly all using the same basic 50s apparatus plus a few tweaks. Yet the tweaks all differ across the branches, so what is taught to all is the same apparatus and every 10 years anew, the tweaks get brushed aside by a new bunch coming in to that territory with the common apparatus and some different tweaks.

Hotelling essentially saw this as what happens with large groups with something to lose: they become conservative. Kahneman noted that the basic apparatus has a logic and a usefulness and that you just dont get the same mileage out of adding a few tweaks, meaning that the whole field is stuck because there is no way to develop a completely different framework within the tribe and complete resistance to rival frameworks from elsewhere.

The ‘classic’ historical solution is then for some small other tribe to do it and take over the whole field. The takeover then happens when the whole tribe is doing badly and a new virulent, aggressive, and most of all successful invading group (where success is measured in a political/money sense) can take over. Yet, mainstream economics is still in rude health, hogging lots of space. So why should it worry about a challenge. And why would it tolerate truly disruptive theorising?

I like to describe the situation by saying 1950s economists insisted people behave like fish (as-if macro). The innovation of the 70s was to insist they are fish (the rational expectations revolution). The innovation of the 00s is that people are biased fish who should be fitted with snorkels (behavioural econ). No-one gets away with suggesting that humans are really not fish at al, that they are something else, and that it is undesirable for humans to be like fish in the first place. Particularly the last bit is still blasphemy.

Mike Pepperday
Mike Pepperday
3 years ago
Reply to  paul frijters

Thanks, Paul. That is quite informative.

Well, I agree how hard it is to budge the establishment. That is the problem with the post hoc statistics too. It’s entrenched and even big name objections as from Marshall, Lewin and Hayek don’t make any impression. Still, chin up; mustn’t give in, you know.

Just a correction. I am not suggesting tweaks. If that is what you got from what I wrote, you misunderstood. I am suggesting coercion and cooperation are of equivalent importance to competition. That is, they are of equivalent importance for economics. Tweaking is not in the picture. Makes it all the harder I guess.

I am not suggesting any alteration to the theorising on competition; I am suggest add-ons. I am saying it is insufficient to calculate the fall of the artillery shell using the theory of gravity. You do need the theory of gravity but you also have to take into account air resistance.

When I responded on this thread, I had not read Nicholas’s post from 10.2.20 (Griffith Review article). It is relevant and I have posted a response to him there.